Long-Term Investing: 155 Years of S&P 500 Holding Periods and What Selling Cost

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-11-07

Long-term investing means buying investments and keeping them for many years, usually ten or more, so that reinvested dividends compound and the bad years get outweighed by the good ones. Its two working parts are time and the discipline not to sell when prices fall.

Most guides to long-term investing state that it works and move on to tips. This page tests the two claims underneath it, that time shrinks the risk of a loss and that staying in beats getting out, on 155 years of monthly S&P 500 data with dividends.

Every start month from January 1871 is counted, so the periods overlap. Returns include reinvested dividends and use Robert Shiller’s monthly data, described below.

How it works

A long hold changes which risk you are carrying. Over one year the main risk is a fall. Over thirty the main risk is that the money grows too slowly to beat rising prices, which is why the table’s inflation column matters as much as the loss column.

Dividends are a large part of the result. Shiller’s monthly dividend figures are added back into the index here; a price-only chart understates what a holder who reinvested actually earned.

Time does not remove falls; it gives them room to be recovered. A twenty-year holder still lived through crashes. What the data shows is that none of them, from any start month, was deep and late enough to leave the whole twenty years below the starting point.

The behavior is the hard part. Every row of the table assumes the holder did nothing for the whole period, through 1929, 1974, 2008 and every scare between.

The original data

What was measured: Robert Shiller’s monthly S&P Composite series, January 1871 to June 2026, the last month with a dividend figure in the file downloaded on 7 Nov 2026. Each month’s total return is the change in price plus one-twelfth of the yearly dividend; after-inflation returns divide by Shiller’s consumer price index. The rebuilt series matched Shiller’s own real total return column to rounding error.

Prices here are monthly averages of daily closes, not month-end closes, which smooths the worst days. Calendar-year data gives slightly different counts; the time horizon page runs the same question on calendar years from 1928.

The typical result hardly changes with length; the extremes do. The median after inflation stays between 6.77% and 8.83% a year at every length; the gap between worst and best shrinks from more than 200 points to about 9.

Since 1950 the loss counts are lower, the inflation counts are not. For starts from January 1950, 2.88% of ten-year holds ended below the start, but 15.79% lost buying power, because of the 1970s.

The longest wait depends on whether inflation is counted. Before inflation, the index with dividends took 184 months, from September 1929 to January 1945, to get back above its high; after inflation that fall was made up by November 1936, because prices dropped in the early 1930s.

After inflation, the longest wait was 153 months, from August 2000 to May 2013. All of the holding periods are in the published file.

What selling in a fall cost

The test: $10,000 in the S&P 500 with dividends from January 1934, the first month with Treasury bill data, to June 2026. A fall counts when the index first closes a month 20% or more below its previous high; it ends when that high is regained. Cash earns the 3-month Treasury bill rate. No taxes or costs.

On monthly averages there were six such falls, starting in June 1962, May 1970, May 1974, November 1987, September 2001 and October 2008.

On daily closes the 2020 and 2022 declines were bear markets, but averaging each month’s prices and adding dividends trims them to 18.92% and 19.26%, so they do not count here; neither does the 19.77% slide from May 1946 to May 1947. The 1937 crash is not a separate fall because the 1929 high had not yet been regained.

Selling and coming back a year later left 60.53% of the stay-invested balance. It helped in 2 of the 6 falls, 1974 and 2001, when the next twelve months were weak. In the other four the market rose between 13.47% and 38.46% in the year the seller sat in cash.

Waiting for the old high left 23.11%. That seller missed every recovery in full: from the October 2008 sale to the August 2012 high, the market gained 57.76% while cash earned 0.41%.

After inflation the gap is the same proportion. In January 1934 dollars, the stay-invested balance was $6.75 million, the one-year seller’s $4.09 million and the old-high seller’s $1.56 million, since prices rose 25.30 times over the period. The month-by-month values are published.

A worked example

Take $10,000 put into the S&P 500 with dividends in January 2000, close to the top of the dot-com market and one of the worst starts in the record.

By February 2003 it was worth $6,131. From that low it got back to $10,000 in April 2006, fell again in 2008, and in January 2010, ten years in, stood at $9,435. A ten-year holder had nothing to show for a decade.

Kept to June 2026, 26.42 years, it was worth $83,818, 8.38% a year. After inflation that is $42,367 in January 2000 dollars, 5.62% a year. In buying power it sat below the starting $10,000 from October 2000 until March 2013.

The lesson is about the length, not the start. Nobody could know in January 2000 that it was a bad month; what changed the outcome was staying for the second half of the period.

What a long-term plan writes down

The figures above describe a whole market, held untouched. A written long-term plan usually turns that into four decisions made in advance:

• What to hold: a spread of companies rather than a few, the point of diversification, with the stock and bond split set by asset allocation.

• What it costs: every year of a thirty-year hold pays the fund’s fee, which is why the expense ratio compounds against the holder.

• How money goes in: a fixed amount on a fixed date, so no single month carries the whole decision.

• What happens in a 20% fall: decided before it starts, since the test above is the cost of deciding during one.

When it fails

It fails when the money is needed before the period is over. A ten-year hold from March 1999 lost 5.92% a year after inflation; anyone who had to spend it in 2009 got the loss, not the recovery.

It fails when the holder sells in the fall. In the 1934 to 2026 test, selling once per fall and waiting a year cost about 40% of the final balance.

It fails when long-term is measured in dollars instead of buying power. 11.05% of ten-year holds since 1871 ended with less buying power than they started with, while showing a gain on paper.

It fails when the holding is not a whole market. These figures are an index of hundreds of companies that replaces the losers; a single stock held for thirty years can go to zero, and nothing here says otherwise.

It fails when the past is read as a promise. This is one country’s market, chosen partly because it did well, and its 1,626 overlapping twenty-year periods contain only seven that do not overlap.

And it fails during withdrawals. A portfolio being spent from cannot simply wait out a fall, which is the problem a withdrawal strategy has to solve.

Questions people ask

How long is long-term investing?

Usually ten years or more. In this data ten years cut the share of losing holds to 2.92%, and twenty years cut it to zero, before inflation.

Is it better to stay invested in a crash?

In this 1934 to 2026 test, yes in four of six falls. Selling and returning a year later ended with 60.53% of the stay-invested balance.

What if I invest at a market peak?

January 2000 is the record’s answer: below the start ten years later, then $83,818 from $10,000 by June 2026.

Can you lose money long-term in stocks?

In the whole S&P 500 with dividends, not over any twenty-year hold since 1871, though one twenty-year hold lost to inflation. Single stocks and shorter holds can and did.

Related

Getting out and back in on a forecast is covered under market timing, and what the falls look like from the inside is the subject of drawdown. The drawdown recovery calculator works out how long a fall of any size needs to recover, and index funds are the usual way to own the whole market these figures describe.

To compare nine ways of investing the same $10,000, see investing strategies.

This page is educational, not financial advice. Test every idea on your own charts before risking money.